John E. Purviance

Properties of FET parameter statistical data bases

Statistical databases are often used to characterize the statistics of a FET. It is shown that a database containing the FET model parameter marginal probability density functions and covariance matrix is not sufficient to describe the FETs S-parameter statistics. This result is important to those developing statistical databases for GaAs FETs. The implications of this work for simulation and CAD are discussed, and a solution to this problem, the truth model, is presented.<<ETX>>

Scalar sinusoidal feedback laws in decentralized control

Sinusoidal, time-varying feedback is employed to eliminate fixed modes in decentralized control systems. Energy and frequency considerations (related to the transmission channel) are used to develop the control law. The technique is demonstrated using a simple two station example.

Statistical design for microwave systems

The authors present a methodology for statistical system design using the commercial simulator OMNISYS. Using two examples, a simple amplifier chain and filter, and a complete satellite receiver system, the application and benefits of statistical system design are demonstrated. The nonlinear characteristics of the system amplifiers and mixers are accounted for. The specification of group delay, signal to noise ratio, and power out are all considered in these statistical designs.<<ETX>>

Yield sensitivity of HEMT circuits to process parameter variations

The authors summarize the use of a graphical tool, yield factor histograms, to study the yield sensitivity of HEMT circuits to process parameter variations. A computer program called SPICENTER is used to incorporate the HEMT statistical physical model with a SPICE circuit model and then to generate the yield factor histograms and yield sensitivities as functions of the process parameters. The authors present the application of these tools to digital and microwave circuits. Two example HEMT circuits, a two-input NOR gate and an inverter chain, illustrate the concepts. Yield sensitivity is presented as yield percent change per parameter percent change. >

A measure for system monitoring and sensor placement

Output uniformity and the sensor placement problem are defined for LTI systems. An index of output uniformity is developed from the model order reduction literature. The usefulness of the index is demonstrated with a pressurized water reactor (PWR) pressurizer model. Areas of present and future research in this area are described.

Determining if a noise corrupted matrix is singular

The problem of determining if a square matrix is singular by examining the matrix corrupted by additive noise with known statistics is addressed. This problem occurs in numerous identification algorithms. The concept of Normed Sense Singularity is introduced to solve this problem. A threshold test is developed which statistically determines Normed Sense Singularity by examining the matrix properties of the singular values and the spectral norm.

Direct structure identification of linear multivariable systems from operating records

A structure identification algorithm for linear multivariable systems is described that permits colored measurement and process noise as well as control inputs. An apparently new canonical form for matrix fraction descriptions (MFD), the Structure Canonical form, is used to develop the algorithm. The identification is made from operating records rather than responses to special inputs, is noniterative and is thus termed direct identification. The algorithm requires testing a noise corrupted matrix for singularity and a threshold test for singularity using singular values is developed. The results extend those of previous studies which consider white measurement and process noise and do not allow control inputs. They can be interpreted as generalizing to the multivariable case partial autocorrelation techniques used in time series analysis.